# AI opportunity decision record
Use fictional or approved information. This is a planning aid, not an investment forecast or release authorization.

## Decision boundary
- Team, problem and next commitment:
- Decision owner and review date:
- Available specialist/reviewer hours:
- Excluded actions:

## Opportunity register
| Candidate | Current process and baseline | Proposed change | Non-AI alternative | Affected people | Evidence: observed / estimated / unknown |
|---|---|---|---|---|---|
| | | | | | |

## Gates, before ranking
| Candidate | Required condition for this scope | Pass / unmet / unknown | Evidence and owner | Next action |
|---|---|---|---|---|
| | | | | |

## Comparable value scenario
- Eligible tasks per period:
- Baseline minutes per accepted task:
- Proposed minutes per accepted task, including review/correction:
- Shared upkeep minutes, excluding any work counted above:
- Net minutes = tasks x (baseline - proposed) - shared upkeep:
- Lower/higher assumptions and adverse scenario:
- Quality-equivalence evidence or unresolved assumption:
- Setup, fees, and other excluded costs:
- Why estimated freed time is not automatically cash savings:

## Capacity and dependencies
- Specialist capacity needed by each candidate:
- Shared dependencies and their cost counted once:
- Capacity retained for rework:
- Plausible assumption change that reverses the ranking:

## Next experiment
- Selected candidate or reason to defer all:
- Why each other candidate is deferred or blocked:
- Bounded task, inputs, permissions and responsible owner:
- Non-AI comparison:
- Evidence to collect, time limit and stop condition:
- Observation that would change this decision:
- What remains unproven before any production pilot:

Source guide: https://learn.aicompetence.org/knowledge/prioritize-ai-opportunities/